Edge-Guided Feature Pyramid Networks: An Edge-Guided Model for Enhanced Small Target Detection

Area of Science:

• Computer Vision
• Artificial Intelligence
• Infrared Imaging

Background:

• Infrared small target detection is crucial for defense applications like precision targeting and monitoring.
• Detecting small drone targets in complex infrared environments is challenging due to target size and imaging distance.

Purpose of the Study:

• To develop a novel model for accurate infrared drone target detection.
• To address information loss during down-sampling in traditional Feature Pyramid Networks (FPNs).

Main Methods:

• Introduced Edge-Guided Feature Pyramid Networks (EG-FPNs) integrating edge characteristics and multi-scale feature fusion.
• Proposed an improved residual block with multi-scale convolution and inter-channel attention.
• Developed layered feature fusion and edge self-fusion modules to enhance edge and multi-scale features.

Main Results:

• EG-FPNs demonstrated improved performance in IoU, nIoU, and F1 metrics compared to existing methods.
• The model effectively captures deep image features while emphasizing edge characteristics.
• Comparative experiments on multiple datasets validated the proposed algorithm's efficacy.

Conclusions:

• EG-FPNs offer a lightweight and effective solution for infrared drone detection.
• The model is well-suited for resource-constrained infrared scenarios.
• The integration of edge features and multi-scale fusion significantly enhances detection accuracy.